The Effect of Tobacco Smoking Differs across Indices of DNA Methylation-Based Aging in an African American Sample: DNA Methylation-Based Indices of Smoking Capture These Effects

Abstract

Smoking is one of the leading preventable causes of morbidity and mortality worldwide, prompting interest in its association with DNA methylation-based measures of biological aging. Considerable progress has been made in developing DNA methylation-based measures that correspond to self-reported smoking status. In addition, assessment of DNA methylation-based aging has been expanded to better capture individual differences in risk for morbidity and mortality. Untested to date, however, is whether smoking is similarly related to older and newer indices of DNA methylation-based aging, and whether DNA methylation-based indices of smoking can be used in lieu of self-reported smoking to examine effects on DNA methylation-based aging measures. In the current investigation we examine mediation of the impact of self-reported cigarette consumption on accelerated, intrinsic DNA methylation-based aging using indices designed to predict chronological aging, phenotypic aging, and mortality risk, as well as a newly developed DNA methylation-based measure of telomere length. Using a sample of 500 African American middle aged smokers and non-smokers, we found that a) self-reported cigarette consumption was associated with accelerated intrinsic DNA methylation-based aging on some but not all DNA methylation-based aging indices, b) for those aging outcomes associated with self-reported cigarette consumption, DNA methylation-based indicators of smoking typically accounted for greater variance than did self-reported cigarette consumption, and c) self-reported cigarette consumption effects on DNA methylation-based aging indices typically were fully mediated by DNA methylation-based indicators of smoking (e.g., PACKYRS from GrimAge; or cg05575921 CpG site). Results suggest that when DNA methylation-based indices of smoking are substituted for self-reported assessments of smoking, they will typically fully reflect the varied impact of cigarette smoking on intrinsic, accelerated DNA methylation-based aging.

Epigenome-450K-wide methylation signatures of active cigarette smoking: The Young Finns Study

“…Several studies on impact of aging on DNA methylation, as summarized by Ciccarone et al, have established age-induced DNA methylation changes as hallmark of aging [ 12 ]. Recent studies such as by Yang et al [ 13 ] and Lei et al [ 14 ] have reported association between cigarette smoking and age-induced DNA methylation changes. Joehanes et al also performed Gene Ontology (GO) [ 15 ] based gene set analysis of the list of statistically significant CpG sites and reported 99 gene sets that broadly included molecular processes such as signal transduction, protein metabolic process, and transcription pathways [ 9 ].…”

Array-Based Epigenetic Aging Indices May Be Racially Biased

“…Such differences have the potential to yield scales that reflect different influences across racial/ethnic groups. For example, in the FHS (White) sample, the DNAm PhenoAge EA index largely loaded on cigarette and alcohol consumption and this was not the case in the FACHS (AA) sample [ 24 , 25 ]. Since the rate of smoking and alcohol consumption are roughly equal in Whites and AAs [ 38 , 39 ], it is unlikely that differences in habits affected the methylation outcomes, suggesting that the DNAm PhenoAge index is operating somewhat differently across ethnic groups.…”

“…The fact that 5 of the CpG sites whose confounding by ethnic specific genetic variation was confirmed in the second set of analyses were also significantly associated with epigenetic smoking in prior studies of the FHS is not unexpected. Recently, we reported a set of analyses that showed that cg05575921 and a GrimAge sub-index (packyears), but not the DNAm PhenoAge EA index strongly predicted smoking status in the FACHS cohort [ 25 ]. The finding that the methylation signal at these loci is affected by ethnic specific variation supports the assertion that confounding may have diminished the association between the DNAm PhenoAge EA index and health behaviour in the FACHS cohort relative to white samples.…”

“…The DNA from these samples were processed via our standard procedures, then interrogated for genome wide methylation using the Infinium MethylationEpic Beadchip by the University of Minnesota Genome Center. Standard sample and probe level quality control were conducted as previously described [ 23 , 25 ]. After quantile normalization, the resulting methylation values were exported as beta values for use in this study.…”

“…In the Framingham Heart Study (FHS) Offspring Cohort, which consists of individuals of European ancestry, 195 of the 513 CpG sites in the DNAm PhenoAge EA Index are significantly associated with methylation at the well-established indicator of smoking, cg05575921 [ 24 ]. However, in AA samples, the Levine PhenoAge Index is only modestly associated with cotinine seropositivity and cg05575921 inferred smoking intensity [ 25 ].…”